1 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Perception  Human vision limitations  Levels of perception  Relative intensity  Illusions of blinking  Illusions of shapes  Illusions of motion  Illusions of depth  Illusion of distortion  Subliminal messages  Pareidolia

2 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Colors and their names YELLOW BLUE ORANGE BLACK RED GREEN PURPLE YELLOW RED ORANGE GREEN BLACK BLUE RED PURPLE GREEN BLUE ORANGE Do not read the words, say the color of each word.

3 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Motivation  Graphics is about communicating with the viewer’s brain  We need to understand how the pixels are perceived and interpreted

4 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac The human eye Light travels through cornea, iris, lens to the retina. The iris contracts: 2 mm (day) -- 8 mm (night) The lens changes shape to achieve focus. The retina translates light into nerve signals. Made of 3 layers of photosensitive cells lining the interior of the eyeball : 1. 1.Ganglion cells (front layer) Bipolar cells, horizontal cells, and amacrine cells Rods and cones (back layer) contain pigments that fire when hit by a photon: release energy and change molecular shape to become less sensitive to light (bleached). The melanin (black pigment) behind the retina helps to chemically restore the light-sensitive visual pigment in the rods and cones after it has been bleached by light.

5 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Rods and cones  120 million rods (1000x more light sensitive than cones, but not to color). More in the periphery. Can see dimmer objects. Better for detecting motion.  6 million cones (see color), concentrated in the macula which includes a 0.3 mm diameter fovea (only cones) –Tristimulus: 64% red, 32% green, 2% blue (light sensitive, out of fovea)

6 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Visual acuity  Normal visual acuity: Ability to resolve a spatial pattern separated by a visual angle of one minute of arc –We would need 360*60=21600 columns of pixels on a panoramic display around the viewer to produce the illusion of a continuous image

7 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Blind spot A small portion of the retina where the optic nerve connects to the brain has no receptors. Close your right eye. With your left eye, look at the L below. Slowly move your head closer or further away from the screen while looking at the L. The R will disappear when your head is approximately 50 cm (20 in) from the screen.

8 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Dominant eye  Hold your thumb in front of you at arm length  Look past it with both eyes  Close the right eye.  If the thumb jumps to the right, they your left eye is dominant.

9 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Stereo  Use inverted binoculars to look at these

10 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Illusion of blinking

11 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Illusion of scintillating

12 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Illusions of motion

13 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Mach band  In 1865, Mach discovered that we perceive an edge (change of intensity) at lines that separate surface regions with identical intensity on both sides, but different intensity derivatives.  It is caused by the lateral inhibition of the receptors in the eyes.

14 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Lateral Inhibition  Consider a light pattern on the retina that changes abruptly from dark to bright  The photoreceptors (green) communicate to the retina output neurons (red dots): –An excitatory input to the corresponding neuron (black lines) –Inhibitory input to the adjacent neurons (blue lines) so that we detect variations with respect to the surrounding intensity (contrast)  The signal in the output neurons (purple line) shows how this helps detect edges and steep gradients.

15 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac More contrast illusions

16 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Herman Grid  A ganglion cell looking at a crossing sees 4 bright patches in the inhibitory neighborhood. When looking at a street, only 2.  The effect vanishes when we look at a crossing, because the fovea has higher resolution (smaller neighborhood)

17 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Variation on contrast illusion  Why do we perceive an X?

18 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Intensity is relative Squares A and B have the same intensity. Illusion is due to colors of neighbors and to the fact that we compensate for the shadow

19 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Contrast illusions enhanced by 3D interpretation

20 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Contrast adaptation

21 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Afterimage Impression on the retina remaining after the initial stimulus is removed. It has colors that are complementary to those of the original image. Look steadily at the cross in the center of the picture to see a green afterimage rotating.

22 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Perceived shapes and motions

23 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Illusion of relative motion  Background rotates  Spokes are fixed  Yet they appear to rotate

24 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Waves

25 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Rotating Snakes

26 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Explanation of motion illusion  “Rotating Snake” by Akiyoshi Kitaoka,   Rotation perceived during eye movements.  On steady fixation the effect vanishes.  Asymmetric luminance steps cause illusory movement  Direction of perceived motion from light to medium

27 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Perceived motion irregularities The two squares move at constant speed. Half the time, no motion cue. Moving objects appear to slow down at low contrast.

28 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Perceived distortions

29 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Scale illusions due to context

30 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Size illusions due to perspective

31 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac More size illusions due to perspective

32 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Illusions due to shading

33 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Perceived depth

34 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Impossible figures

35 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Faces - Pareidolia See a face?

36 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Subliminal messages

37 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Resources      

38 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Additional Reading

39 Georgia Tech, IIC, GVU, 2006 MAGIC Lab Rossignac Examples of questions  If we wanted to build a panoramic theater where viewers would not see the pixels, how many colums of pixels would we need? How many pixels total (assuming 90 degrees vertical field of view)?  Explain why one portion of the image may appear darker than another of exactly the same color.  Explain why we may perceive an edge on a smooth surface.  Explain why spots appear at the crossings of the Herman grid.  Why do Kitaoka’s snakes rotate?  Why do we see faces in strange places? What is the name of this tendency?